The Uses of Game Theory: Against a Purely Positive Interpretation of Game Theoretic Results
by
Abhijit V. Banerjee
Department of Economics, Massachusetts Institute of Technology

This paper argues that there should limits on how far we should go in terms of giving a positive interpretation to Game Theoretic results. It is clear that people do behave strategically and that Game Theory can be useful in predicting the choices they make, but there is no presumption that the actual game they play is, in any useful sense, optimal. Specifically, in a significant fraction of the cases where people are observed to be playing some kind of game, there is probably another game (i.e., a mechanism) that they could be playing that would make all of them better off, but either they have not thought about it or, having thought about it, have not yet arrived at the optimum. Therefore, when we observe people making certain rather inefficient-looking choices, we should not presume that there must be an economic reason why they could not do better.

A related issue is how one should make use of the revelation principle. In other words, should we assume that while people may use an indirect mechanism what they have in mind is always the appropriate direct mechanism? Or is it possible that the ïrrelevant" aspects of the mechanisms (how we label choices, etc.) make a difference with respect to how people behave?

I will begin by illustrating these two points using two examples, one from the literature on risk sharing in village economies and the other from the literature on corruption. An implication of the fact that people use mechanisms that are not the ones that game theorists would have them to use, is that there is scope for innovation—in other words, the normative role of Game Theory needs to taken more seriously. I will conclude by discussing two examples, one from Bangladesh and one from India, where there have been recent innovations in how people carry out certain routine transactions.

Date received: July 5, 2000

Copyright © 2000 by the author(s). The author(s) of this document and the organizers of the conference have granted their consent to include this abstract in Atlas Conferences Inc. Document # cafl-19.